Comment on your cell selection criteria.
Peak voltage may indicate something, but mid charge voltage shows more about the entire cell, internal interconnects, cell resistance, etc. A partial dump from full charge and under load voltage might be better to sort cells. AC impedance testing is used by the industry to measure internal resistance but voltage with a medium load vs time (2 amps for 5 minutes for example) would be a quick way to weed out poor cells. If a new cell gives a voltage outside the group distribution at 70 to 80% charge, it is an outlier.
I have the ability to graph/log all of the above data. Including internal resistance. Good suggestions. How about this. I will do a full charge/discharge analysis of the cells (at 1c). I will then post the data, and we can all pick the cells as a "team". That way every ones concerns are addressed. LiFePO4's mid level charge characteristics are really quirky, compared to "standard" Li-Ion cells. They really do jump around a lot....another option would be to take the cells to "storage voltage" (70%) and allow them to sit for a few days. Whichever cells are closest to the average at that level, could make the cut?